Slashdot Mirror
Google Donating Bandwidth and Servers to Wikipedia
Armstr0ng writes "According to Dirson's blog, Google plans to help Wikipedia by donating bandwidth and servers to handle part of their increasing load. In fact, there's an official page of Google's proposal to host some of the content of the Wikimedia projects."
It usually is a 10 minute process to login to Wikipedia and call up an edit page. Not to mention that this might help with all of the Slashdottings Wikipedia's servers have to survive;)
If my answers frighten you, stop asking scary questions.
Plenty more people would do so if there were an official mirroring system for Wikipedia, which there is not. All kinds of people would be tripping over themselves to lend bandwidth and servers to them if there were such as process, and IIRC they've gotten offers before from universities and such.
If Google wants to help out, I don't see why they should be get any kind of special access. The ball is not in Google's court, but in Wikipedia's.
(No disrespect to Vibber and the guys keeping the servers at Wikipedia HQ online; they're doing god's work. But the site would probably be a lot more stable with an army of official mirrors than with a single, monolithic server farm.)
Hopefully both of them will disappear. Right now, the only reason they're included is that the internal search engine was overloading the servers.
"They redundantly repeated themselves over and over again incessantly without end ad infinitum" -- ibid.
No plans for Google ads. We do send search to either Google or Yahoo now when the database servers are under an uncomfortable amount of load. I'm usually the person making that decision and I decide solely based on balancing server load and reduced service.
Instead of static caches we're using Squid caches which get updated automatically when the content changes. We're looking to place more of these in other places which use significant amounts of bandwidth or are far from Florida in response time terms.
We're aware of the risk of excessive dependence on one donor and are looking to avoid it. We're entirely happy to talk with other companies who want to share in being seen to be helping something obviously good, limited only by the suitability of the offers for our needs. I don't know what the Wikimedia Foundation board would say but personally I'm entirely happy to accept hosting from Yahoo, Microsoft, AOL or any other significant player. In my personal opinion, neutral means just that - including neutral toward all companies in the business.
If any competitor is concerned about Google getting undue benefit or prominence, the solution is simple enough: talk with us as well.
We have offers of various sorts in Europe (though more, particularly from major carriers with excellent peering arrangements, would be very welcome) and the US. We don't yet have a substantial offer in Asia and that's a major hole I'd personally love to see filled. We're very popular in Japan and a location with good peering in Japan would be good for service there.
By this time next year I'd like to see 2-6 major remote sites with database slaves and apache web servers, capable of taking over as master if there's a failure in Florida, plus 10-20+ remote Squid caching sites. A massive amount of work (and donations) required to get that done.
We've already been blocked from China ourselves on several occasions. I've little doubt that it'll happen again and in other places as well from time to time.
It's absolutely certain that we have some unknown, uncorrected copyright infringement, offensive content, politically incorrect items for various parts of the world and assorted other things some or many people find undesirable. If the chairman of the board or president of the country is making headlines worldwide for some indiscretion, expect it to be in the article. Nobody who is unduly concerned about such things should consider offering hosting - we can't guarantee the absence of such content, just that we will try to be neutral.
We're not only interested in hosting and bandwidth. I'm particularly interested in high performance disk drives or systems, high capacity RAM modules (database servers like RAM but 32GB of ECC costs $11,000...) or whole high power database servers. To give some idea, I'm thinking in terms of three quad Opterons with 32GB of RAM and 12-16 15,000 RPM SCSI drives to keep up with demand for just the English language encyclopedia project over the next 6-9 months.
No part of this post should be taken as representing the official views of the Wikimedia Foundation or any members of its board. It is, of course, blatant soliciting for donations, as you'd expect from the guy who does much of the capacity planning...:)